INNERTRON SENTINEL
Neuromorphic metacognition · Independent oversight of artificial intelligence
For eight years, Dr Rory Lewis and his research team have been developing an independent, brain-inspired neuromorphic system to help protect people from potentially harmful AI advice and behavior. Rather than relying on an AI system to police itself, INNERTRON SENTINEL investigates a separate architecture designed to recognize when human reconsideration may be warranted.
The approach draws on human metacognition: our ability to examine our own thoughts without automatically accepting them. By independently evaluating interactions with large language models, the proposed system would provide a graded concern signal when advice appears uncertain, inconsistent, misleading, or potentially maladaptive. Its purpose is to support human judgment, not replace it.
The materials below provide technical detail, preliminary results, and prior research supporting the proposed Google DeepMind research effort and related NSF proposals. Select any document preview to open its PDF in a new tab.
1. NSF IIDS research proposal
This proposal develops a unified mathematical framework linking neuromorphic dynamics, evolving connectivity, and fuzzy and rough set representations of uncertain information. It examines how evidence accumulates, how connections develop, and when these processes can produce stable, meaningful signals for human reconsideration.
The one-page IIDS summary summarizes the entire 18-page IIDS proposal. Read it first for the central research question, proposed mathematical advances, and broader impacts.
2. FET research proposal
A brain-inspired architecture for independent oversight of artificial intelligence. The full proposal describes prior neuromorphic hardware research, preliminary results, mathematical foundations, the proposed architecture, and its experimental evaluation.
Start with the one-page summary. It summarizes the entire 18-page FET proposal, including its overview, intellectual merit, and broader impacts.
3. Prior research · 10th World Conference on Smart Trends in Systems, Security, and Sustainability (WorldS4)
Dr. Lewis presented The Case for Neuromorphic Sentience to Control Large-Scale Artificial Intelligence at WorldS4 in London, UK, July 28–30, 2026. The paper presents the theoretical foundation and prior research underlying the proposed independent neuromorphic architecture.
